#πβsuno-school
1 messages Β· Page 83 of 1
I 100% agree that you pay for a service you should be able to use the service. That said, this doesn't last long at all
yeah no it stated happing 20 min ago
im not waiting
It rarely lasts 20 minutes
and it hapend agian a few hours ago
@cerulean merlin Please understand NOBODY is able to "fix this" in the way you want. Improvements are ongoing. It's not a switch that can be flipped. π
its been happending lots of times for the last few days
Ok. This was happening before Timbaland endorsed Suno. After that it got smashed and worse. They have to have the time to update upgrade and everything else. Gotta have the funds to pay for everything.
It normally only lasts for a few minutes
...to everyone. You're preaching to the choir. We've all been affected by the outages.
It is the database getting over loaded
If I was Timbaland, I would be donating some funds to upgrade the cloud service...lol
I hate that I am still finishing my Bachelors, not saying they would hire me, but I guarantee you the first thing I would be looking at is what to do about the current database. What it is using which type of database and the amount of incoming connections and a dedicated DNS. Suno isn't going to get smaller if they keep innovating. If they were a public company I'd buy stock
Make bank baby!
What I can say is that the database is only partial. Since whatever happened yesterday only came from the EU direction that is something totally different and yet it acted the same. Couldn't connect. That I blame on the reauthorization being looked at too often. That basically creates its own DoS attack.
one of the teams are helping now and they said they will send it over to the devs
im dm
You're a saint, Mike. π
(second guess being DAViDπ€)
I'm glad you got an answer. Sounds like they are more personal than I thought. Go Suno for caring about your users!
Then that happens...haha
Hey app you don't say anything when people are actually using profanity...lol
lol your right at first it was mike π€« but its up now for me thanks everyone for helping and thanks to you mainly u been helping me the most for the last few days i really appreciate it
π€·π»ββοΈ
i was trying to ask how to prevent it from cutting a second off the end of songs randomly.
i cropped something and it was perfect for the first 5 listens and then now when i go back to it its missing the last note and sounds bad
You can't prevent it but you can prompt it to be more consistent
and the original isn't in my trash so i can't go back and fix it
Wait so you heard the whole song and then it cut off after?
https://suno.com/@chatterwolfsunosongs π π€«
Can you share it, without the embed
listened to it multiple times on repeat before i went to publish
idk how to make it not embed
how in the hell do I write a instruction for say (heavy synth, and bass) and it not sing the line guys please.. its doing my head in..lol
(parentheses) are for backing vocals
[metatags] are for localized style prompts
AH... you legend.. thank you..hahah
I see it is extended. You sure this is the version that you were listening to?
i mostly want to know why stuff isnt appearing in my trash any more too. i could just redo the crop but for some reason suddenly now is when stuff doesn't go to trash
yes, every other version should be in my trash rn lol
ill just do replace section on the end and hopefully get something similar
It is strange for the Audio to actually change just on the way it gets generated. If you heard it once on the full song then it exists. The entire track was generated...it is very odd to hear a cut track after simply leaving your page and coming back
Thank you for your patience π« I know I come across as grumpy sometimes -- I didn't mean to sound grumpy or to make you feel like I wasn't hearing YOUR story when me & my buddy SixStringToker said "oh yeah, that happens to everyone" 
I think of it like Suno is a second-hand car that we're all sharing -- so we're just letting you know "sometimes you need to kick it really hard to get it to start, and be careful that the steering wheel doesn't get stuck after too sharp of a right turn. Also, it'll stall if you come to a complete halt, so try to run as many red lights as you safely can." π
I'm glad that you're here with us, and I know that soon enough you'll be the community-expert letting new Suno-users what kind of quirks to expect when they try to find the same sucesses that you have π β€οΈ
Persona is using Lyrics prompt from its creation, not the actual one and creating almost the same song over and over again!
So are we having analogy contests now? Because I got some Leslie Neilson ones that are just awesome, offensive, but awesome...lol
This happens often. When you use a persona it is like taking the original song and using it to influence the current song. Because of that and being in beta, it can poke through sometimes.
If the lyrics are recognized by the original then it will probably poke through
Leslie, Neilsson, are you being Frank with us ?
Is there way to simply use extend to replay on instrument on another without keeping the original audio ?
If I understand, like using "Get Whole Song" but only getting the extensions without the original clip that's being extended from?
I'm getting the same and for some reason I can't generate a song beyond 3:15 seconds and no matter how short the song is, it keeps cutting off. I've been trying to use the extension function to fix it but it's just not flowing correctly, it's very frustrating
Are you using personas?
Another thing with the extension feature is that I'll but it as instrumental but it'll just randomly generate words still
Lol....I like my sex like I like my basketball, one on one with as little dribbling as possible.
Not a bug, a feature -- lemme explain π
When you make a persona from a song, it takes a chunk of it and tucks it aside so that when you make a new song using that persona, it listens to that chunk first to "remember" what it's supposed to sound like. Because v3.5 has a 4m limit, it seems like your persona is using a 45s clip to remember waht to sound like
Because of this, you're only left with 3m15s for the remaining song -- a way around this would be to crop out the BEST sounding part of your song & create a persona based on that, like an exceptionally powerful vocal section, or something demonstrating a duet, or a song with really good instrumental tones -- whatever you're trying to capture π
If you only give Suno a 20s clip to work with, that's the longest the persona can feasibly use as context, plus this way you ENSURE that it only uses the part that you wanted, instead of capturing sections that didn't reflect why you fell in love with the song in the first place
roger, Roger.
Thank you! I'll give it a try
Yes sir remove the original clip since it's my low quality sketches on acoustic and it's ruining everything
I used to think personas were garbage until I realized I wasn't making creative use of them -- I wish you the same successes I've found with it π
(btw. I don't know if that's your actual cat, but seeing a kitten in a witches hat is exactly what I didn't know I needed this morning, so thank you)
And stop calling me Shirley
If it's just a matter of getting the song without the original, you could "Get Whole Song" and crop out the original, but bear in mind that Extension uses the existing audio to continue from -- if there's something about the audio you don't like, you'd need to make changes somehow before using Extend -- potentially you could Cover it to "reroll" the song, but that's a less controllable method, even if you use the same style prompts.
It's just really low quality iPhone recordings I like to sketch melodies and chords and then extend from there
I think I wasn't using it right. I'll try your advice from the other thread and see if it works. Thanks for the help!
And it is not my cat but I use it every year around Halloween. It's my favorite cat picture on the internet lol
Not that you were using it incorrectly, but maybe hadn't found ways to use it to it's full potential! π
If it means anything, if I ever see that picture again, I'm going to think "hey, that's @cinder sparrow's cat-picture", so in a way it IS yours π Have a great one!
is there a way to keep what suno has created but have it redo it with a lyric change ? as I like the music but would like the vocal to change a bit?
Cover works
There's the Replace Section feature that lets you choose a specific segment to regenerate -- not a perfect 1:1 method of getting the exact same instrumentals, but it's the best simple & generative option that doesn't require external audio editing
ok . thank you

Popcorn? So early?
Why not? Itβs already 10:30 am here
Ok so here is something strange, all of my songs are coming out with a British accent.
While I have no issue with the accent, it doesn't go in every style
interesting, do you have any tags or styles that seem to be influencing it?
has it been happening more consistently to your songs recently?
A little yesterday and everything today
hmm, that's strange
I think Suno rotates for real
who really knows at this point
True
The song sounds good but the British accent can make American slang sound crazy
oh for sure
Even though it makes no sense. I think because i am using shortened Rap version of words that it is interpreting it as a British accent.
that could be it, though it's hard to say what's exactly causing it
Yep. That is the one constant...lol
More hellenic death/folk/thrash metal stuff
https://suno.com/song/aec30965-e605-4672-a951-2c4ae3a2c8d8
greek folk metal, technical death metal, groovy speed metal, high-energy crossover thrash, grindcore, aulos, classical song. Listen and make your own with Suno.
Well it's Beer Hall Putsch day... and Downton Abbey had an episode about it, and Downton Abbey is a British show... soooo.... here we are...!
The fact that what you are saying makes sense is crazy...lol
Trying to generate covers and keeps kicking off generation but erroring. Got a stack of unplayable songs showing on IOS app but not in web browser.. this happening to anyone else?
Morning everyone
Maybe suno has moved the server to England. π
morning!
Happy National Chocolate Milkshake Day
A good subject for an AI song.
Yea. I wouldn't doubt it. Say convo, g'day and brekky and it'll be Australian, I guess.
How can I get it to do nonhuman robot steven hawking sounding vocals it will not listen to anything I say even with a bunch of exclude human voice tags and synthetic robot monotone vocals nothing works
As an off-guess... try "Speak and Spell" it might know the toy voice.
I was just thinking text to speech or auto narrator speak and spell is good idea i will try that now
Make up something, yea, like "robo-narrator"? Something might work.
Wonder if Suno knows "Dr Sbaitso" ?
You can put prompts reminiscent of the 80s, like "hypnagogic" to see if it nudges it further.
Yea it's still like 50 50 human robot rn I'll try that I have so many things I've tried already but none of these and it's getting closer
I'd test with you, but I'm working. You're on your own. Good luck and have fun.
Re-order the same prompts. That can re-weigh the priorities of each.
Sorry if I say something you already know. Better twice than never.
What do you use to produce very powerful bits?
No u r helping alot I am closer it's working like partially then human cuts through but nah you are good thank you
I have recreated some of my songs using Personas and these songs are often cut short for some reason (around the 3:30 mark). Anybody else see this behavior?
True
It shouldn't take 100 credits worth of Garbage to get robot vocals it's pretty ridiculous but this works 25% of the time or so Id say out of like 50 songs I got like maybe 4 decent ones. [((hypnagogic automated synthetic
speech emulator robotic machine androgynous robo filtered speech synthesizer nonhuman speak and spell text to speech auto narrator robo narrator vocals))]
Sometimes less is more π€·ββοΈ
I tried
Each individually
As well
I also tried Watson ai vocals, Steven hawking vocals, all kinds of stuff
Not directly suggesting that using them individually is guaranteed to work, just noting that stacking that many in a row might work detrimentally to how one might imagine.
True
will try reducing it some
could be like, the concept of a double negative im doing by redundancy
I generally feel like the more specific of a sound Iβm trying to create, the less likely I am to achieve it solely through the use of complex prompting.
Top-of-my-head approach, maybe you could generate the exact robotic vocals that you want with an external tool (Watson AI as you mentioned, if thatβs something that could achieve the vocal tone), then reupload it into Suno to extend from β that way it would have an easier time continuing the provided vocals.
have not thought of that, might be the way to go
so if i get a good audio vocal robot track i just upload audio and then extend or remix it to get the music behind it?
Definitely a more direct approach β yeah, a couple ways to handle it
- uploading JUST the robot vocals and hoping that Suno recalls that vocal pattern for your inputted lyrics
Or, likely more effectively,
- downloading an instrumental song and adding the robot vocals overtop that externally, then reuploading it to Suno to extend from
Iβve seen some crazy experiments, where people have overlaid 2 separate & distinct tracks & overlaid them (creating a complete garbled mess to human ears), but Suno was able to recognize specific patterns blended within that clip to create an extension that somehow made sense of the chaos, while still continuing those tones & style.
Similarly, you might not necessarily need to get the robot voice sounding perfect overtop the initial generation just to get things started β you can always try to clean things up afterward with a cover or replace section π
right on thank you very much
Permanently deleted tracks apparently do not actually get deleted at all, at least not if they've ever been extended from. I can use "Get Whole Song" on extensionsnssjd fdhffgjjgf
hahah
THAT was Rachael
I thought you were headbutting the keyboard out of anger
Turns out youβve contracted Rachel to headbutt the keyboard for you
....as I was saying before having my keyboard stolen, permanently deleted tracks apparently do not actually get deleted at all, at least not if they've ever been extended from. I can use "Get Whole Song" on extensions of deleted tracks and it will pull in the bit that was permanently deleted.
It's probably all relative to our own library. Permanently deleted from our personal library, but still on Suno's server for them to do with as they will. 
so i found steven hawking emulator site text to speech, upload that audio into suno, then i extend but i change the extend from 57 sec to start at 1 second, and it worked
Nice β you can use Crop or external tools to clip out any unseemly sections at the very beginning, but thatβs awesome, dude!
πͺ Nice work on that one
yea it actually sounds even crazier cool than i intended
Hello everyone, I have an issue with Suno, I'm saaad.... Since Yesterday when I tried to create a song (After putting my lyrics, theme and title), it loads into nothingness and after my song is not create, someone know how to solve this ? Sorry for my english
How to use negative prompt in suno? I was searching on youtube and google but couldn't find any info on it.
It works the opposite to the "style of music" prompt -- enter things that you DON'T want to see.
If you're making a rock song and keep getting a piano solo, add "piano" to the negative prompt section.
I can't find the negative option. Mybe I have to enable it somewhere?
I love it because I feel like you can go a little more wild in the negative prompt section to steer the generations away from a specific result without worrying about drowning out your desired style prompts.
For instance, if you say reggae, disco and it strongly focuses on disco, an older (and arguably effective, however blunt) method is to stack the ignored prompt, like reggae, reggae, reggae, disco which is more likely to give you reggae at the cost of how well balanced or respected the other style prompts would be considered.
Underneath the "instrumental" toggle, you need to toggle that feature on
Oh. The excludes style?
damn. hahaha. I was looking for something that says negative prompt
"Exclude these styles"
Jesus.. lol
Thank you so much!
I guess I got mixed up with stable diffusion too much
Reasonable oversight π
After months of trying to find creative ways to suggest music without specific instruments, tones, styles, etc, it's nice to have a spot to "yell at" suno in a productive manner π
guitar solo, frantic, shredding, excited, sporadic, flighty, manic, turbulent, tremolo-picked, guitar-sweep being included in the negative prompting of most songs I make tells a story in itself π
I should try those out. My English is not that good. Sometime I can't describe the things I don't wanna hear π
Like last time, the loud humming sub base sound and I thought it's called crackle or something else
It's nice that Suno "vibes" with whatever we say, so it's less important to know the technically accurate terms, but I understand that sort of familiarity acts as a sort of "roadmap" in knowing what else to try.
loud bass, hum, crackle, too loud, booming, turtles, people who say they hate pineapple on pizza without trying it first -- obviously not "the best" prompts, not "the correct prompts" -- but ESPECIALLY in the negative prompting, you can literally do nothing wrong.
The worst case in the example I provided is that you won't hear people who hate Hawaiian pizza. π
Those are good negative prompt. I think those are good for my instruments. Thanks a lot!
that's the turtle doing in there? j/k
maybe put some Ketchup and Mustard sauce in too π
Heck -- even if those words make ZERO difference in the negative prompts, it's another way of making them uniquely yours. If "turtles" isn't associated with music in a way that would negatively impact the song, it's like an "easter egg" surprise for your fans to find π
π Anything is possible with AI
I remember you mention, less is more, now I am getting pretty good instrumental song with just a simple style prompt. No need to list any instruments in the section descriptions.
There can be times when Suno doesn't respond as well, or when songs take longer to generate -- possibly even getting stuck and failing.
If this is happening for longer than 30m, and it is impacting EVERY song that you make, regardless of lyrics or style prompts used, it might be a good idea to contact support@suno.com so the Official Support team can help πͺ
Using many style prompts is a very fun creative experiment, but I find that it can be VERY helpful, like you are, starting off with simple & few prompts -- you REALLY gain an understanding of what you can expect in your own "little corner" of Suno in this way.
I wonder if suno server store the cache of our style and keep looping similar creation, if we keep on creating. I constantly refresh my browser.
Sometime it feel like after 2 hours of continues creation. The songs starting to sound the same. Could be my ears?
Does Suno not follow prompts anymore LOL? I ask for Drum and Bass and get Rock.. .I ask for female vocals, i get a guy
and if I DO get Drum and Bass, its super mellow and slow and more like lofi. Suno...she is broken
Previously I would have said 100% Suno does not work like this -- I asked a staff member recently and they weren't sure themselves. A dev has been quoted as saying "I wish it worked like that" -- still, I have seen MANY more instances that would prove otherwise, since it's unfair to give preferential credence to the instances that confirm the theory & discount the majority of times that it wouldn't have impacted the generations.
I'd be curious if it continues to happen with simple prompting as well, as there might be some overlap in the prompts used if you're hopping from one style to another
Pass me the lofi. I love it.
i can confirm this. after 3k songs, i'm basically repeating what's been done already. I have a good amount of songs that sound identical in structure and overall beat and chords. It's like its reset to "0" again.
I think it's good. But possible over lapse. Just curious about it.
for me more importantly now is this style prompt issue. I really do not understand why this is happening. How can you make a rock song with a prompt that says " Dark Core, Deep bass synth, liquid drum and bass"
i wouldn't care if when you logged it as a bug you got the credits back, but to my knowledge you do not.
Just to make sure I conveyed the message I intended to -- I have ~9k songs and haven't noticed a resounding "sameness" to them, looking past the formulaic nature of music in general, but dont send me down that rabbit hole π
I personally don't believe that Suno uses user-history or thumbs up/down for immediate feedback in generation results, imho
that's crazy. I have a bunch of my metal stuff that is structually the same. same tempo changes as well.
Hum.. about the thumbs up/down thing. I wonder if it's just training the AI overall when we click it. or it's just for individual account that telling suno don't make this kinda style/sounds/instrument again. So I am not sure if we should remove the songs that are thumbs down?
I am afraid that he can't remember the context from previous creation for the new songs if we delete it.
afaik, it provides training data for them to use when developing future models, but I wasn't under the impression that it impacts songs between generations when using the same style prompts -- otherwise it would be absolutely USELESS for people to "reuse prompt" because it would be entirely different for everyone based on their history π€·ββοΈ
My conspiracy theory is gaining ground. Lol.
Hey. One thing we may at least remind ourselves is that melodies and patterns DO get repeated, as they wouldn't be trendy song patterns or hooks and be part of western music theory if they didnt. After hearing hundreds of AI generated songs, we start hearing the patterns and think something is amiss, but they've been there all along.
π I'm still in the other camp since nothing definitive has been shown, and the only dev comment still speaks against it, but I'm softening my absolute refusal
Hey everyone. I have a rather stupid question: Is it somehow possible to change email addresses from an account where I used google to register? Because I want to change my email to a safer one.
Not stupid at all -- not anything that you'd be able to manage on your own, but perhaps you could email billing@suno.com to ask if there's anything they can do to change the email associated with your account πͺ
are the servers down or something i just cant generate it, before this error message it was coming up with the songs but just loading and not competing them.
this was happening yesterday. I think they were rolling out a fix or something (maybe just changing servers) . If they were rolling out something it would be based on location (West coast goes down first, then middle of US, then east coast) as an example.
I would also recommend, at the moment, to use a Password Manager to generate complex passwords for your current email if you feel it is insecure.
There are several out there but I believe the most popular were 1Password and BitWarden
so theres nothing i can do?
probably not as it's not on your end.
The thing is, I used my "trash" google email to register, because back then I through I would gave Suno only few tries just to mess a little. But nowadays I generated many tracks and even bought a Pro plan, and so I would like to keep my account safe.
However, thanks for the email! I will email them if they can change my email.
could also be like Midjourney where you used up all your spots in generation and have to wait. what plan are you on?
maybe vpn?
Probably wait 10 minutes. If it still gives that message after that, contact support@suno.com
each plan has limits on how many you can do at once.
ive been waiting for 30 minutes now tbh so i should contact them i assume
Ah, yes, contact them ASAP
About 3 hrs to end of the workday and weekend time
Hopefully they don't have a backlog and see your message about it
I use BitWarden, I love it.
Version 4 I hope when doing covers, it does not spit out the same song and charges us for it
Hopefully! (Just remember that you do get like 200 covers for free #π’βannouncements message. There is a visual glitch that shows the cost being deducted but refreshing the page after the generation is complete will show the correct amount)
same with Persona, to an extent
Version 4 will probably be a major upgrade considering how well it is doing covers and persona when it comes to Vocals. That along with fixing the drift however that is done without breaking it into segments.
creating new songs doesnt work atm, it just loading and loading and loading...
I'm having the same problem right now....
Ok so we need to wait π
Anyone using accessibility tools like screen readers: The model dropdown is currently unusable. The previous, older UI was better because it was a standard dropdown, but now it doesn't work at all. Can you try to get the old dropdown back or do any of you know a workaround?
but seriously, has it been confirmed if personas is a different model or not?
Be sure to post this in bugs channel, but I'll try and highlight this to Suno as well https://discord.com/channels/1069381916492562582/1082775596767117312
It uses v3.5
then how does it usually sound much better? π
Even Suno's Bark can output impressively clear sounding music, without changing the model, if you change how you use the model.
Because personas take a clip of the song that you made it from -- presumably you've already deemed the song it uses to be subjectively good, so you're a fan of the tracks that result from it.
Instead of using the same random prompts to hope for a track that "rolled the dice" the same way, you're forcing Suno to listen to what you like, and demanding that it "do more of that, but different"
Imo -- it's essentially a v3.5 extension that doesn't seamlessly blend with the context clip that you've provided, so the songs can have their own intros without feeling like they're a direct continuation of the existing song -- same as all other extensions, "better quality input == better quality output"
Is there desktop computer software that can reliably just tell you the key and chord progression in any audio file? Actually I should just look for solutions on github, command line would be better anyway. Want to do get a sense of diversity of outputs, and even if had a musical ear, would not be able to listen to a million songs.
Free, or good?
Doesn't matter as long as it's mostly accurate. Able to take in a batch of songs, or be automated, is probably what I mostly want.
I haven't used them myself, but I chatted with someone who spoke REALLY highly of Melodyne -- I think that's what the youtube channel "Now You Ruined It" uses
https://mixedinkey.com/learn-more/ also gets mentioned ,but I'm less familiar
Is anyone else having problems with the ''Get Whole Song'' function?
nothing is happening here when I click on it.
Oh I challenge you all to tell me how I did the speed rap. Though the answer is easier than you might think. I think I've told you about this in DM once @charred merlin so if I did, you disqualified from challenge! I will reveal how though. https://suno.com/song/c9d60643-eff2-4353-a4b2-f24c4fd77467 (https://suno.com/song/8478a152-b4b1-4965-bd6b-8918f1c9726d is the original that I didn't go experimental on)
You might need to refresh the page -- it should show up as the most recent the library track -- make sure you don't have "Liked" toggled on though π
I'm abstaining from this challenge, but I'm glad that worked so well! π I'll keep that idea tucked away as a "good one"
it's also not just how it was structured. The hardest part isn't the prompt
cause you can do the prompt and end up with, nothing. I also temporarily created a new genre when JvC (original) blew up for awhile
so many people just typed in rapidcore rap and thought that would give them the magic potion
Linebreaks and empty used to be super super influencial in Suno V0. To the point where almost any song that just had a lot of lyrics, like a lot of raw text, always wanted to come out as rap. (There was no style prompt.)
yeah that ain't it though and some speed rapping just won't work, regardless of your prompt or technique cause it doesn't flow and creates a break in the speed itself. That's why speed rapping tends to lend itself to certain consonant, adverb and syllable structure and why chopping was created as a way to string together words that can be multiple other words and connect them and all that. Speed rap and chop isn't the same thing though. I learned that after the fact, lol
There are many things you can use. https://multimedia.easeus.com/bpm-key-finder/
Chord AI on both android and iPhone, FL Studio on Mobile or PC
Is there a way to download a video with the current name and picture if I changed it?
Regenerate Video option should update the video, but it may take a couple minutes before the download link gives you the new version
It keeps using the very first name and picture
Wait wdym by download link?
doesn't always do that. If it refuses, you can crop the entire song after changing everything and then regen the crop before you link anywhere to it (if there's an existing embed it might interfere with the regeneration of the embed itself)
if there's an embed anywhere on discord, anywhere at all
?
Embed
don't work
nothing appears in the library even when updating the page, simply the get whole song function doesn't work...
If there is an embed link on discord active, it can interfere with regeneration and you have no control if someone has linked it
that's why adding / works sometimes
if nobody has linked it with that already
Never heard that before -- is this something that you know for a fact?
I know that the moment something is linked, I can't regen it
or should I say an embed is embedded
I don't have a sub going so I can't say if it works by deleting it
I haven't run into that, either. Just tested it with a track I shared on discord. I changed the cover, regened, then downloaded the video and it had the new cover.
What I HAVE seen is discord caching the videos. So if I were to repost that here, it'd likely show the old cover, but that's a discord thing.
you can download it yes but can you link the embed again with the new video?
Toker -- I was literally about to ask if you'd test that out with me, but sounds like you already played both roles
π
thats the thing I'm saying
it does regen but the embed doesn't and I deduce it's cause it's cached somewhere on discord
Yeah, discord will cache the old video. There's probably not to much Suno can do about that.
Okay, we might have been talking about different things. Discord DOES cache, but someone watching the video doesn't impact whether or not Suno's regen feature works
See? Old cover. lol
not actively watching, but I mean if it's linked. Just linked. That means it's gonna stay cached
that's why I suspect it strongly cause if you add a / at the end, you can sometimes make the regen work
until you've linked it
Discord isn't the same as Suno though -- the video is updated, but Discord's cached version doesn't
You can post the link without the / elsewhere and it will still show the updated video
Adding the / is just a way of sharing the same song from a different link that hasn't been cached yet
I know but...we were talking about embeds cause why regen videos otherwise?
still has the old cover once the vid starts
I might have lost the point of all of this -- you would regen a video to hide extension points, to change the displayed lyrics or image
To change the image, to censor certain words, to hide clutter, to get a better structure of the lyrics to make them readable. To add copyrights
to my lyrics, not the song
how come my coins vanished? I had about 2200. I did cancel my sub a few months ago. Shouldn't I still have them since I did pay for them
kinda feel jipped
Think with the paid for credits you still need active sub if I recall that right
Anyone know if v4 is still 4min limit? Or 3:20 on cover/persona
Monthly credits or paid-extra credits?
Monthly credits
ok found it under subscription page
Personas each use a segment of their original song to help give context & keep them sounding like "the same artist", so that persona uses 40s
Thanks @zinc cove
no probs!
Yeah, was just wonder if any of this was changing in v4
And well aware this is more of a hardware/memory limit then a technology limit
afaik It's meant to ensure that Suno can balance the server load as they grow so quickly -- easier to keep an idea of how much stress on the network there'll be this way, but it definitely catches people offguard if you aren't expecting it -- I know it's ABSOLUTELY me trying to make the best of a bad situation, but there's a helpful guide about how a user makes use of the last week-or-so during the billing cycle to go crazy with unused credits that'll expire https://discord.com/channels/1069381916492562582/1285616051190632510
Heh as a former copywriter...I will say, it's mostly about making money out of people not using the service to its full capacity. Just like gyms
bind yourself for 12+ months, you probably won't go for more than 3 months
This actually does raise a question, lets say i had a song that i really liked the vocals on, and cut it down to just a 10 or 20 second clip and used that as a persona, would the persona generate longer on that persona?
I've commented that EXACT thing over the past couple days, and people have found great success with it from early reports -- instead of giving it a 4m song, and hoping that it knows which segment to use that reflects why you fell in love with the song in the first place, crop out the section that best represents what you want and then use that for the persona π -- a guitar solo with really good mid tones, a duet, the perfect metal vocalist, etc
https://x.com/sunomusic/status/1854961069300613176 ok so I think some of the v3.5 generations are already v4 and they are a/b testing it. #πβsuno-school message
Cool, ill give that a shot, and can you create a persona from an uploaded file? Like upload myself singing?
The red x react mean no i guess?? Lol, i undsrstand why they wouldnt let it, copywrite issues and all
Well from above it seems like it just samples it and uses it as kind of like how a lora/embed would work on image gen.
Although i would like seed control
Not in the technical sense of something that would allow you to recreate the same song multiple times using the same prompt, but in the sense that you'd use a seed crystal to grow a larger crystal that uses the same patterns
yeah I don't mean the seed recreating the same thing (unless it allows you to recreate everything, which it doesn't) but in the same way a seed in image generation will function as a guide even if you alter the prompt to deliver a similar "style" to it
I'm sure the back end has steps and everything and such that we don't get to interact with
Not sure, but in image gen, if i keep the same seed and change even 1 character in the prompt, the images look nothing alike, but with embeds and loras i can get the same characters, styles, etc
Wait in image Gen you get different images with the same seed? I am asking because it is the exact opposite for me and I think everyone else. If I use the exact same seed with no variation I can change the prompt a decent amount and still get the look and stance of the person or whatever was in the previous. As long as I use the same model of course.
Yeah, with same seed, changing a single thing in prompt gives me different image
Interesting, I notice that I forget to set the seed to random very often because I keep getting the same image over and over. Well the same background and basic everything else maybe an outfit will change or something. That is crazy, and is this in Stable Diffusion?
Yeah, running forge on either sdxl or flux
Huh..,that is nuts.
that's very weird
Can you use personas to remake your old songs?
forge might be doing something else
BRB. I'll explain persona in contrast to seed in a second
Yes and no, it will some times reuse parts kf original song, but not consistantly
The coolest result i had from a person, i was using a metal song as the persona, but was doing kind of a slow, more accoustic intro to that song, and where i had a [break] in the song, it dopped into the heavy riff of the metal song, then back out to the accoustic sound when vocals picked back up, then at the end of the song, rather then fade out like normal, it gave me an awesome drum into and heavy hit that i was able to crop and time perfectly with the start of the other song
So. I don't know if anyone else caught, this or realized it.... And im probably looking too far into this so I could be wrong. But I took a listen to the v4 teaser song again and one lyric stood out to me "feels like I've been waiting for a lifetime, looking forward to today" I don't know if that's just part of the song, but what if that's a clue that v4 is dropping even earlier than we think? Possibly even tonight?
Doesn't work
https://suno.com/song/71a68d8c-be62-46e7-bd53-569087f4ad11 I do these things for fun now, horror songs ^^ I got the INK itch I also make up genres cause it creates cool sounds...and I have no persona's
no sub
example:
`[Thriller Metal Verse]
Dear Dieary.
Last night I burnt something I loathed.
I tore it down and shattered it apart.
Her screams drowned out the room.
Now sheβs ashes along with her home.
[Instrumental Interlude]
[Thriller Metal Riff]
[Horrorshow Punk Chorus]
Nobody sees me.
Iβm invisible to the world.
Since I blend into the canvas.
I may as well paint it full with blood.
Nobody sees me.
Iβm invisible to the world.
Since I blend into the canvas.
I may as well paint it full with blood.
[Instrumental Interlude]
[Evil Darksynth Solo]`
created a very modern 80s sound
and technically there are slasher horror villains in PG13 movies now so. It works ^^ But feel free to tell me if it doesn't and I'll delete the link
https://suno.com/song/2fd62930-df59-484f-93ff-2c27b5461cbe this was the halloween challenge where I got a Muse sounding vocalist. Def want to persona this if I ever resub
Slashercore, Horror Metal, Fearsynth, e-string quartet, Male Voice, Vcal Fry, fight or flight, adrenaline, visceral song. Listen and make your own with Suno.
I would like the images to be generated a little better...
I re did all mine in stable diffusion with a trained embed of me
I made my cover from SD + Clip Studio Paint
got my template and then modified it, redrew some stuff
added the title and the "stars" that are anagrams of slasher villains
Yeah, that is a great work without any of hallucinations!
I have an old SD model I trained using blocks, unet lanes, redirecting lane trafic and siphoning very specific data from various models
cause I didn't have the vram to train an actual
https://civitai.com/models/6209/dark-gemini-v3 this one if anyone is interested in an old 1.5er. Can do up to 1280x720 80% of the time without assistance from regions etc
my job requires me to have 4 4090's
any idea how to clean up the vocals, i have no problem working in another program outside suno if you have suggestions
I'm happy my SD model has very positive rating and I think nothing but 5/5 star reviews
yes I brag π
what do you mean clean up? like remove them?
like in the actual song, the vocals have a kind of a tin sound to them.
thought about making an sdxl model?
i've played with one song after stem split, but not even sure where to start to clean them up
do you have a wav file of it? the song
cause mp3 is too low quality download to clean up
oh yeah, i get that, im trying to work on flux, and even thats hell, the models themself take 20gb vram, try to do anything cool with them and your out of luck
and yeah, i did the wavs
im just not sure where to start, i have fl studio and i know that pretty well but haven't done anything with vocals so a little clueless
if you have them I can take a look tomorrow in studio one 7
it's midnight now so about to go to bed
see if there's a vocal there to save
lol yeah, ill send you links
to the full one, not stem split
I will stem split it on my own in SO7
better stem split quality
cinematic, deep, dark, emotional, depressing, bipolar, melancholy, angry, Nuclear, Nu metal, slow, black and white, song. Listen and make your own with Suno.
i would really like to do something will all these things i've wrote, just not sure how to clean them up so any advice is useful
I can't download it though, you need to download the wav and send it to me
I'm still learning, so I'm not a professional...but I consider taking on projects like these to learn more
yes it does, lol
and youve got nothing to lose anyway π
lol right, i think its funny that with suno, if i do a metal song its fine, but anything with clean vocals is like.. meh...
https://suno.com/@fallingforever I mean you can check my songs. There's some clean vocals https://suno.com/song/0ae76403-d3b1-40bd-bc91-9baeac5b0b90
Emo or Gothic oriented in various genres, sometimes typical metal/rock variants with various electronic/synth. Sometimes pop, All lyrics written by me Β©
lascivious, avidity, sensual, submissive, sultry, Trancecore, EDM, bedroom pop, pop, Female Vocals, 110 bpm, Major IV song. Listen and make your own with Suno.
ffs
that was 10 seconds!
lol
there problem solved
yeah, female vocals are easier with clean i've noticed
listening to sick six sick, i have a song called the sin you might like, afraid to post it here. a few songs i've done i've been on the line of allowed and not
Acoustic, Gothic Metal, Goth Symphonic Metal, Pained & Sorrowful male voice, Baritone, tragic, sad, Melancholic song. Listen and make your own with Suno.
no wait
that's without persona
yeah, that vocal sound, that weird.. i don't even know how to describe it, but thats what I am trying to fix. you really hear it in the first 2 lines of that song
yeah it does.
and occasionally comes back
for a second or two
https://suno.com/song/7cd6b489-7e53-4651-93bd-e2c209efa57a here then...from Sept 4, long before personas etc and even inpaint, crop etc
Agonizing, Tragic, Gothic Metal, Neoclassical Metal, Male Singer, Powerful voice, Baritone B2-C3 range, resonance, 70bpm song. Listen and make your own with Suno.
like this song, the vocals came out good, but it was darker tone https://suno.com/song/5d574a38-5261-46d3-a0a3-f976fce5e5cf
Male Vocals,deep,dark,Angry,depressing, bipolar,Nuclear,black and white,low tone,nu metal,progressive metal,complex song. Listen and make your own with Suno.
the intro is supposed to be distorted btw
Hi Guys. IM having some trouble getting Suno to replace the lyric in a section. any ideas?
the entire song was an homage to The Crow (1994)
yeah, hope its not homage to the new one lol
its litrally not singing what Is written
in my cases, if its being stubborn, i've done a cover
or go back further if you are doing a extend
im doing a "replace" edit
never tried that, so not sure, i've always just covered or extended
π€π€
I can DM you a link to my horror song playlist (when I get the impulse I sit down and write lyrics for 10-20 minutes and then play around on Suno without a sub. I don't use AI for my lyrics, even if it would improve some of them)
you can use inpaint and if you tick off the time box you can extend them however long you want. Just edit the song first to include only the new lyrics you want and then add them in the new prompt
yes please, and yeah, I don't use ai for lyrics either, not sure ai could even write half my lyrics,
suno is constantly telling me my lyrics need "reviewed"
i did a song for a friend of mine, about being kidnapped (she's way into bdsm).. suno did not want to do that song.
I have remade numerous songs with persona I call them persona mix even. The vocals and harmonies are noticeably better.
With persona vs. Seed. This is how it is functioning now, I have no idea if this is what was intended. If the seed was being held then any change to the song would not make much difference, except in your case oddly. Still as it stands Persona instead makes an actual copy of the original song under the new song, so to compare it would be more like influenced generation on SD like CPDS or Canny, stuff like that.
That is why the original song can poke through if you don't change enough lyrics
Im not to sure what you mean. I only want it to sing the lyrics that are already there. but its singing the first part of the chorus then changing the last have to the first chorus. if that makes sense?..hahah
that's Suno for ya
hahah
if you have a sub you can select that area in specific and just untick the time box and let it run
oooh i see what you are seeing, I have this happen to, a lot when I am doing "nu metal" it always wants to repeat the first or second verse later in the song, even if I give it like 7 or 8 verses
I got around that by not labeling verses, i just label the chorus and it fixed this for me
oh ok
I make like 10-12 syllable verses and stack them to create a larger verse (at least as far as the listener knows)
that way it won't jump back a verse etc and I can make the verse as long as I want
and interject some instrumental interludes
https://suno.com/song/81244169-73c0-4c0a-8a54-0697fbeb7002 when MMORPGs are taken too far
Gothic Rock, Gothic Darksynth, E-String Quartet, Fantasy Journey, Darkest Fate, Male Voice, Female Voice song. Listen and make your own with Suno.
yeah, thats what I started doing, basicly suno just sees it as one long wall of text with a few [chorus] tags in it
also if you understand consants syllables etc for speed rapping. I can tell you a working method
Rapidcore Rap, Rap Chopping, Hip Hop, Male Voice, Rapid Flow, Confrontational, Adenaline Fueled, Inner Battle, 140 bpm song. Listen and make your own with Suno.
20 seconds or so in youll hear
Hey, I try to generate a song but it does not work. I have tried different browsers, using the app, using on my phone and on my pc. But nothing seems to work
what do you mean? does it give an error or what?
with rapping, for me, it was to stop using periods. no more . just use , huge difference
Epic Rock Americana Alternative, Anthemic, Soprano Female Vocals, dubstep, violin, glitch dubstep, Soaring Female Vocals song. Listen and make your own with Suno.
It just times out, nothing else discribed
still slower π
I wrote the lyrics for that a long time ago and updated them for Suno. Suno is the first to ever bring them to life. The whole point was to be an almost constant drumming.
but good
Slower...oh my bad...lol.
Epic Rap Dubstep, Speed Rap, Rapid Fire Female Vocals, distorted guitar, Deep Bass, Sidechain Compression, hard bass song. Listen and make your own with Suno.
Picture needs work I just finished it today
When I post it in showcase I explain one great thing about speed rap is you get to write about speed rapping in the rap at least once...lol
Easy content
https://suno.com/song/c9d60643-eff2-4353-a4b2-f24c4fd77467 check this then π
Epic Speed Rap, Rapid Fire Rap Metal, Rap Chopping, Male Voice, Vocal Fry, Visceral Anger, Terror, Nihilistic, Resigned song. Listen and make your own with Suno.
my experimental version
I like Female Vocals. I would say going with Latino male would probably haul ass
about 25ish seconds it the first one starts
K
For mine it is the chorus and the end. I wasn't happy with the end, I could have kept going but I was over the song already...haha
the redux is just me messiing around, the original was the real one
a lot of it is more or less nonsensical to the story the original song was telling in redux
Jamming this now
there's a method to getting the speed rap down, including both chopping, syllables, consonants etc for the actual writing. Then a small trick
the last speed rap verse I just decided to go for it
Yep. First off your song jams, I love it. Had I did speed metal I would have gotten that level of constant speed. I literally removed all commas from the song. Still yours obviously took more time than a day. Great job
never the time thats an issue for me, its the fact that I can burn through 3,000 credits on one song
`[Final Verse]
Kill_it_then_eat_it_never_feed_it_donβt_believe_it_just_leave_it_donβt_free_it_itβs_the_devil_itβs_not_clever_so_forever_remember_to_sever_the_head_of_every_dread_ful_feelin_thru_the_shadows_in_the_gallows_where_the_hallows_creep_in_the_shad_oh_so_late_the_hour_am_swallowed_by_the_hollow_never_follow_where_the_needy_n_the_greedy_lose_the_gamble_itβs_too_shallow_fail_in_the_straight_n_narrow_instead_the_arrow_hits_me_every_hour_so_I_cower_til_Iβm_empowered_climbin_towers_for_hour_son_loudes_ton_set_won_it_in_the_middle_o_the_run_nits_versus_them_adnes_sadnes_sudden_lie_sick_ing_ames_en_the_shame_o_the_painless_resurrection_icing_the_cake_uprising_versus_our_placing!
Always gonna be Jester Versus Clown!
All_the_haters_are_just_waiters_cya_later_my_master_traitor_put_some_bait_on!`
[Verse 2] killin_n_winnin_never_endin_never_bendin_always_spendin_endless_energy Fueling me!. Makin_noradrenaline_n_dopamine_all_over_me_like_amphetamine_in_a_blender_never_tire_in_this_mire_soarin_higher_with_this_vitamin_not RITALIN! You_feelin_me_Iβm_in_the_ceiliin_CLAWS_out_n_everythin_fightin_my_DEMONS_they_never_tire_Iβm_the_entire_purifier_collider_of_the_rider_from_then_tire_under_world_on_fire_scape_of_tape_it_lend_in_a_hook_for_the_croo_educated_over_ate_the_muse_crated_I_waited_never_sated! Related_ill_fated_mating_sedating_going_dating_guess_itβs_okay_then_nobody_cared_when_cosby_did_the_slow_he_needs_medicine_too_Bro! Not_just_his_date_to_be_(REDACTED)_you_really_wanna_feel_me_n_tell_me_why_I_canβt_say_(REDACTED)!
Ah OK, yes because I took a couple hours to do mine. The thing that took me the longest was being swapping over to phonetic spelling, but it takes up more space and I didn't get the content I wanted. This is a more quality song of mine considering it actually tells a story. https://suno.com/song/52f4d3a4-6ed2-477e-b517-57e21492441c
Epic Orchestral Rap, Aggressive, High-energy, Male Vocals, Motivational, Fast-paced, Gritty, Layered, Trap, glitch song. Listen and make your own with Suno.
basically, chop syllables when needed to create several words out of one and make it flow through it, no line breaks, no spaces, no commas, no exclamation, no question marks. Uninterrupted flow of letters, just going with underscores
but you have to put the words together right or it will stop
so the most important part is done writing the lyrics
Yes Bone Thugs taught us this and 36
now I wanna try an actual rap.. thanks guys. lol
I'm not a rapper though or that into it. I had written rap like once when I made Jester VS Clown
Krazy and Bizzy were kings of syllable chopping
digging that sysmai
once in my whole life pretty much
yeh nice one
too bad Suno is bad at bending vowels and syllables and shifting consonants
I have been called a wordsmith many times but i am not a poet and therefore not a lyricist
so we can't eminem it
Rap God or Godzilla yea
I dunno if I would be called a wordsmith but I'm a lyricist. Then I'm learning production, so I'm a songwriter in training
He completely enunciated every word
I am getting so much garbage out of suno last few days. I guess the prompts i've been using don't work anymore? also...anyone else notice the audio quality isn't as good? lots of artifacts and "underwater" sounding music.
You mean the flanger washout sound
https://suno.com/song/b6a98940-fd59-4bf4-afe8-d885998ece12 this was my first rap song ever
100 Beats Per Minute, Rap, Hip Hop, Electronic Music, Cyberpunk, Tempo song. Listen and make your own with Suno.
This line came out awesome man! --->"as wordsmiths we stay gettin it, fittin n splittin syllables, spittin written miracles, like fundamental lyrical minerals,"
though it has more trap based sound but the lyrics be rap
yep...it is SOOOO in my face now
I talk a lot. So that is the story. Tons of words but they aren't poetry and not necessarily written worthy. So, many have called me a wordsmith since I have convinced many things many times...lol
The best I do to combat that is break up sections and use Tempo changes
I'd love to get Suno V4 early access. They gave me V2 access a year ago for being one of the first subscribers to annual pro.
not even a rap fan but I like this, my rap extends to eminem and nf mostly.
so just do one section at a time?
Not really, first start with Tempo changes. The sound you hear is actually called drift it is AI leftovers on long runs.
However, if you change Tempo or make silence or a break of some type it tends to cut it down
Not all the way, but a good amount
suno needs a fav button, lol
you mean like adding these inbetween lyrics
[Tempo Change]
[Syncopated Rhythm]
if you want to get into rap metal and stuff. Falling in Reverse, Prequel, Popular Monster and then Watch the World Burn in that order
So I do that and then when it builds up too much then I go ahead and extend from that point
Yup. Even though they aren't anything to do with tempo
Prequel is epic rap + operatic, Popular Monster is rap metal + rock, Watch The World Burn is battle rap, rap metal, chopping, speed rapping and heavy metal
oh I love falling in reverse
doing something like Watch the World Burn is gonna be hard
not really
I mean yes
but not really on Suno
but an actual song like that? Hell yeah difficult, I hope I get that good some day
You get what I'm saying though, the point of the Tempo Change is not even to make the change but to try and get AI to check itself
voices in my head probably my favorite falling in reverse song, with zombified and watch the world burn right behind that.. the last resort cover was respectable.. but.. odd too
ahhh I see. yea now I do.
I've tried, what came out was...horrible at best
I know, I watched that video
and the first time he heard it he was vibing it
but FIR also did Im Not A Vampire Revamped and The Drug In Me Is You, both same as Last Resort in style
Ronnie wants to stretch his singing talents and musical talents and that bastard is so good
to me it was like disturbed doing sound of silence, to far from their typical sound for me to really love it. unlike johnny cash doing hurt and trent just is like, dude, thats your song now.
FIR changed the way they make music though, now they do what they feel like
Bah
and Ronnie's vocal range has gone from sounding the same in every single song to incredible range
lol wtf was that?
that was woody not liking what was typed
The app that doesn't work right
your font was to big?
Did it actually delete my post
yes
Dude. It makes errors at best
yeah you just gotta deal with it though
Point is I love Sound of Silence by David. Unbelievable
do you guys number each lyric section at the side?
I first heard it when watching Blacklist
i used to until I stopped labeling verses
No
I see.
oh btw the reason I know that Suno reads speed rap like speed rap is cause I did this and then deleted it. To test if Eminems would speed rap, it did
Fill_'em_with_the_venom_and_eliminate_'em,_other_words,_I_Minute_Maid_'em_I_don't_wanna_hurt_'em,_but_I_did,_I'm_in_a_fit_of_rage,_I'm_murderin'_again,_nobody_will_evade_I'm_finna_kill_'em_and_dump_all_their_fuckin'_bodies_in_the_lake_Obliteratin'_everything,_incinerate_a_renegade,_I'm_here_to_make_anybody_who_want_it_with_the_pen_afraid_But_don't_nobody_want_it,_but_they're_gonna_get_it_anyway_'Cause_I'm_beginnin'_to_feel_like_I'm_mentally_ill,_I'm_Attila,_kill_or_be_killed,_I'm_a_killer_bee,_the_vanilla_gorilla_You_bringin'_the_killer_within_me_outta_me,_you_don't_wanna_be_the_enemy_of_the_demo_who_entered_me_And_be_on_the_receivin'_end_of_me,_what_stupidity_it'd_be,_every_bit_of_me's_the_epitome_of_a_spitter_When_I'm_in_the_vicinity,_mother****r,_you_better_duck,_or_you_finna_be_dead_the_minute_you_run_into_me_A_hunnid_percent_of_you_is_a_fifth_of_a_percent_of_me,_I'm_'bout_to_fuckin'_finish_you,_bitch,_I'm_unfadable_You_wanna_battle,_I'm_available,_I'm_blowin'_up_like_an_inflatable_I'm_undebatable,_I'm_unavoidable,_I'm_unevadable_I'm_on_the_toilet_bowl,_I_got_a_trailer_full_of_money,_and_I'm_paid_in_full,_I'm_not_afraid_to_pull_a
I needed confirmation that it understood consonants etc
(that's the end of Godzilla)
would anyone happen to know what the issue is?
What do you mean it doesn't work?
and idk, i just always liked bobaflex's cover of tsos better
download ccleaner
ccleaner?
clear windows cache etc
It neever generates and just times out
Nothing
k let me try
https://www.ccleaner.com/ get the free version, it's enough
Don't download CCleaner
why?
real
you could also try difference browser, I use firefox and don't really have any issues, but have had issues in chrome
I've tried multiple browsers and the app
just hasnt worked
I have been fixing computers my whole life, and CCleaner has been the only tool I used that ever deleted more than needed and actually created issues
tried my computer and my phone
the app is still beta
I'm having a bit of an issue trying to get suno to change the way a chorus is sung and when trying to do the replace section it just mimics the old one
There are others but that is not the issue
but then don't click in the "warning" boxes
trying to change lyrics or just how its sung?
just how it's sung
god this is f'n terrible
https://suno.com/song/5a3904ea-c5d2-4712-82e8-f10c51fb14fc
Rapidcore Rap, Death Metal, Metalcore, Epic Rap Dubstep, Epic Speed Rap, Rapid Fire Rap Metal, song. Listen and make your own with Suno.
I am saying I don't recommend telling someone to download that program when you don't know how much they know. Especially if the issue is persisting across devices
tried bending vowels and consonants, changing out syllables to similar sounding ones?
tried a cover? I've used covers to fix things like that before, but changes are if its timing or singing the chorus way out from what you want, it may never fix, you could try changing how the lyrics are written to see if it "forces" it to change it.
but it will literally warn them, like "this could result in windows becoming unstable or non functioning"
I'm sorry we got derailed what is happening?
no worries, basically, it never generates and it times out
I cleared cache and still not working
either way, cleaning up temp cache etc is something you can try
I've tried the app, my phone, pc, and multiple browsers
Are you from the US?
I cleared all cache
yuh
changing the words slightly worked
windows web cache, temporary files, history, cookies (save the cookies you need first, back them up)
I am asking because multiple people have come in from the EU with the same issue today
if its an issue across devices, its your internet
Hello Friends, exist a way to improve audio quality in suno?
it could just be Suno sucking too
hold up, found the problem.
i had some lyrics it didnt like
elaborate
rofl!
teehee
welcome to my world
I was going to suggest changing networks
what does that mean
@glad valley hello
roll on floor laughing.. is that not used anymore, am I that old?
yeah, I just put some lyrics that werent acccepted in it. but it didnt say that was the reason. Usually it does
You are very old, yes
That has been the most common thing
well thanks for the help anyway man
It's funny "You're dead" is forbidden from being used in lyrics and is called a "producer tag", like what?...I can change it to your dead and it's fine. Or I can say "flay you with my finger nails, frying you like bacon n eggs"
and it wont mind
so not like "You're dead" is so bad, I wonder why it's a producer tag?
People have gotten irritated when they were told it was because of their lyrics so I tried to steer away from that first...lol
I have some pretty offensive lyrics out there and none has ever not worked
I dont post them on discord
I think I'm going to stop for today before i put my fist thru my computer. LOL. Suno is sucking hard this week., it's SOOOO off the mark on song styles.
but they're still there
I've not heard this particular producer tag but there have been instances of Suno generating bits of sound that are very close to existing producer tags. I think they want to avoid people deliberately triggering that. But like I said, I have never heard that specific tag. It's a bit annoying - you can get around it by mis-spelling it but phonetically making it sound the same.
suno's lyric censoring is weird.. can have a line talking about eating ass in one song, but not smacking an ass in another because of the "tone" of the song
and a lot of times adding a random comma fixes the issue
I tried making a song the other day with a persona and got a message saying it was against the content moderation policy .. then I used the exact same lyrics, but without the persona and just a style prompt and it worked fine.
And the words weren't even that bad..
try typing in several ones like "heavy metal symphonic emo punk rock hardcore fusion" and then you can begin a verse etc with [heavy metal verse] then [metal symphonic verse], [punk rock verse], [symphonic emo punk chorus], [punk rock hardcore bridge] or a better structured cluster of genre words that can be assembled in different strings
old school thrash metal song. Listen and make your own with Suno.
and you can get some fascinating stuff
The song in question
i'll give it a shot
omg, most offensive song ever
Right? π©π©π€―
just come up with whatever order can create the most kind of genres
it probably thought you were talking about pooping on people?
Right but it generated fine as soon as I stopped trying to use the personal I had created
That's what's weird
what, that's stupid, just change your lyrics, or spell the word in a way it cant detect
what was the other song about, did it look at the lyrics of both and be like, yeah.. no...
slightly ...and i mean SLIGHTLY...better
https://suno.com/song/82ce19c4-2d8e-4711-8389-e8f74967d34c
Death Metal, Metalcore, Epic Speed Rap, Rapid Fire Rap Metal, song. Listen and make your own with Suno.
If yall don't mind, if yall want to keep posting songs and stuff that is all good but it is becoming enough to flood the tech channel, would ya'll mind removing the embed when you do post one please.
how remove embed?
gotcha. thanks
Yes, I am unsure as to why, but many come in steaming already. Since they sometimes think that we work for Suno
haha, peak
Ok, maybe I'm not explaining this right .. I created a persona from another song that had generated fine and had no offensive words at all - not even "poop" haha .. then I punched in the lyrics for "everybody poops" and got the content warning. Then, I kept the exact same lyrics, poop and everything, took the persona off the prompt [so it wasn't using the persona any more], filled in the style with "old school thrash metal" and got the song I linked, no problem. It only refused to generate when I tried to use the persona. Everything else was exactly the same
should send them a terrible remake of their lyrics and say this will work
You can add <...> around your link before you post it that stops the embed from ever happening
ok, you all inspired me, going to try a rap now
let me know once you're done
Lol. I am not going to tell them how to circumvent Suno's rules in the technical channel...haha
it will be awhile, never try to write rhyming that close, lol
here is an example of making up a genre by using two separate ones or adding an adjective or verb in front of it (Symphonic Avant-Core, Sanatorium Darksynth, Sanctum Metal)
https://suno.com/song/214232f3-4d04-4223-974d-7bdf6e8693b5
Naw, freestyle. I do it all the time with my friends
You from the south?
which resulted in the refrain "Am I a cheater? (Death!)" with the death part coming with a synth sound
That's pretty good!
potato patato
I've done one freestyle, wasn't really a rap song though, just a very angry metal song
I'm from the south, and in my younger days I used to do a lot of freestyling and I'm a white boy...lol
thanks, was more of an experiment π I don't have a lot of songs on my page...but they're either pop for the sake of showing people how easy it is to write pop...and the rest more aggressive
It's so fun, you can get some real gems
REAL
@worn flicker hope you don't mind, going to take a lot of inspiration from tip of the cap for this first attempt, in terms of pacing, syllable setup, etc
This was how I began the instrumentals for Eternal Twilight Reimagined (https://suno.com/song/2a5b7099-f8f1-4e19-98dd-d8c0d9e6311c) which was then cropped to like 30 seconds, followed by a extension at which point the below switched to lyrics.
[#: [Genre: Acoustic, Gothic Metal/Gothic Rock]] [#: [Sub-Genre: Cello Darksynth]] [#: [Styles: Expressive Instrumentals, Melodic Intensity, Vocal Side-Chain]] [#: [Tonic: C Minor]] [#: [Tune: Gothic Ballad, Haunting Voice, Church Reverb, Piano Melody]] [#: [Meter: 3/4]] [#: [Tuning: F#-F#-G-G-A-A ]] [#: [Instruments: Acoustic Guitar, Light Distortion Electric Guitar, base guitar, drums, string quartet]] [#: [Arrangement: Verse, Pre-Chorus, Chorus, Post-Chorus, Verse 2, Pre-Chorus, Chorus, Post-Chorus, Bridge, Final Chorus]] [#: [Composition: Sorrow, Desperate Longong, Resignation, Surrender to Darkness, Yielding Depression]] [#: [Production: Gothic, Melancholy, Aggression]] [#: [Vocal Octave: Baritone]] [#: [Vocal Style: Deep, Subtle Vocal Fry, Cautious Grit, Resonating Sorrow]] [#: [Play Styles: Acoustic Guitar with Reverb, Emotionally heart-wrenching Gothic Electric Guitar Riff, Melodic Drum Ballad, Base Guitar, String Quartet]]
I never finished it though, cause I ran out of subtime and cba to mess around with my invite credits when v4 is around the corner
or maybe it was more like 15 seconds actually
I think I'll remove the vocals and sing them myself at some point
Everyone needs inspiration. No problem.
Don't forget you can create a lot of the song in metatags like that
if you build a starting track with just that and no lyrics and then use it as an extension or cover. You can get the "feel" you want
i've had a few things close to a rap style doing numetal, but never like full on rap so, new to me.
Sure, if you look at my profile I do a lot of Female Vocals and mixed Genres. I love Overlapping lyrics like Broadway, but Suno is no where near being able to run even two sets of lyrics at the same time. A decent amount of my songs I either have to change or force them to go a certain direction because my normal rhyming scheme prompts rap.
What do you mean?
cant run two sets of lyrics?
Correct, can't run two sets of lyrics at the same time
What do you mean?
i've been experimenting with a duet with a friend of mine, My verses in more 5 syllable rhymes, hers is more story with 12 syllable rhymes. its doing okay with that, the biggest trouble is half way through the song it gets confused on who is supposed to sing what and it starts mixing up the lyrics
Listen to a Broadway song, or even the cartoon Encanto, Overlapping lyrics and vocals that sing the same song. Stacked up to six deep.
When they split apart and come back together it is incredible
so people singing over each other or taking turns?
Well Suno can do that with personas
maybe do a cover? and the split out hte vocal tracks and lay them over outside suno?
https://suno.com/song/81244169-73c0-4c0a-8a54-0697fbeb7002 well it can easily do back and forth, I'm leaving the embed for easy listening for now.
Gothic Rock, Gothic Darksynth, E-String Quartet, Fantasy Journey, Darkest Fate, Male Voice, Female Voice song. Listen and make your own with Suno.
Yes, I am not saying I can't do it...lol. I am saying that Suno can't
I know how to do the Duet mid lyric swap
lol yeah, gotcha, i've had some things close to this when using a "chorus" or "choir" tag, but not consistant enough to be usable
I am saying that Suno can't run two sets of lyrics simultaneously. So it definitely can't do 6 stacked
Swapping is just walking it onto the duet
You show me and I'll call myself stupid right here and right now
I forget where the song is I did it in but I know it WAS at least possible
i'll find it tomorrow
it's 2am and I've been up 50 hours
I'm waiting to fall asleep
No problem. I am always willing to learn. I just can tell you that the way Suno builds audio files doesn't even separate then instrument from the Vocals. It will not run an entire Verse stacked on an entire verse at the same time for any amount of time. If so then I haven't seen or heard it. Unless it is uploaded
I know
Oh, so what am I being shown then...lol
but you can run background vocal prompt, and then a regular one
Tomorrow
and it should sing the backgrounds in the background
Yes, that I know
or vice versa
(...), {...}
the background ones won't sound as focused though
I said vice versa cause I can't remember which one comes first. It's hard to think right now
I have ran numerous duets and others like it. Swapping mid lyrics. One or two are public.
What i am talking about is very unique and amazing
What are the supposed improvements coming w/ v4? I can't find anything online.
that they're not talking about it is actually concerning me a bit, that they only show a audio file
Buffy the Vampire Slayer, Joss did a great musical, if you have seen it then you have seen a version. Of what I mean
I hate that episode
and I despise Whedon
He's a good writer, but a terrible person
not great...good, not great
Only the #π’βannouncements post so far
Oh I go with great
all his stories end up the same more or less
Look at the first Avengers
which wasn't even fully written by him
i loved that episode, always liked spikes songs
It is the Giles and Tara part that I adore
Might be a conversation for #π£οΈβchill-chat?
The reprise
It has been like this for a while.
We can cool it
not ever sure how we got down this path.
If I resub, I should probably over this with lyrics https://suno.com/song/69f06558-b944-4e37-8137-26ba6f68ab57
oh, doing duets and stuff, thats how we got here
Yes, we got here honestly...lol
I can see -- I popped in to see if anyone needed help, and tech chat is jumping!
I gladly share most of my secrets
sorry, tech questions lead to othr convo's. we will chill
including that you can type in guitar tabs
This is a actual tech discussion though, I was saying suno can't run two sets of lyrics at the same time
do tell
I have, for a few hours
Hopeful that some of the grain in vocals will be reduced. Also hope we can recreate sections longer than 30 sec. That would be game changing.
Your Underscore information is new to me, thank you for that
np ^^
brb reboot
https://suno.com/song/faea8cd4-d05c-407c-90d3-779e40305638 enjoy this for now, feel free to extend if you like the sound
Yea there is not any info about v4 there wasn't even an announcement about its development from what I read, just the knowledge that they are obviously not going to stop...lol
Before today I mean
π€v3.5 was huge. Hoping for the same. An upscale feature would be lit too.
Just going off persona and cover, this should be something to get excited about
By that I mean the obvious vocal quality difference and belting ability
Vibrato, chorus harmonies happening when they should and not the whole chorus. It took me a lot of messing around to get that to happen before cover and persona
well they've been beta testing some of it on 3.5
stealthily
Right, persona and cover
which is why it breaks so much these days
no I mean things you have no control over
they change the model itself with new training data
breaking prompts that always worked before or screwing up lyrics that always worked before, hallucinating more
jumping around
Yes, you can tell that by the rotating issues.
We not dummies in here...lol. that is why we are in here...lol
the beta test model should be separate and probably the only one available to free users
so in exchange for using the service for free, they test it
well no you don't have to submit your paying customers to beta test
you can split it into 3.5 and "beta"
I should say LLMs and other natural language models
One off topic sentence: that got my crawl for the longest time, I lost it when I saw Steam start selling betas...I'm like what...we are doing them a favor.....
I was so proud of myself when I turned my SD 1.5 model into pure natural language so much that it got used as a foundation for prompt processing for all big models around
I beta tested for a long time, and my days are the ones of people getting paid to test games...lol. They really figured out how to work people.
I never let my kids buy a beta
Perhaps this is plausible now with them opening up to International and Android users.
Ok no more off topic. You sleep as you said. Screens make you stay awake. I'm gonna work on a song. Nice chat
they wanted its ability for natural language and the diverse art styles. It would understand "Powerful goddess with swirling vortex of dark oozing energy, wearing bone armor and crowned by bones growing from her skull, pale skin and yellow eyes, onyx colored veins creeping across her body, mist covered necropolis in background" etc etc. Instead of "1woman, bone armor, pale, somethng_with_hair etc etc" stupid stuff
and it could hold weapons, it could have characters battle and stab each other, it almost always had five digits
always 2 arms 2 legs
I know it's off topic but just wanted to brag about it
It is impressive to get a one shot with all correct digits and anatomy in 1.5 without Loras so, I acknowledge your brag. Now we finish...lol
Yes, thank you ;D
I wonder what LLM that Suno uses
I mean the LLM is the front end
I think it may use a GPT
Has this been stated somewhere, or conjecture?
As one of the alpha testers, this is literally the first I'm hearing about separate models being a/b tested.
considering pretty much everyone I know that uses Suno suddenly has their identical prompts fail. Suddenly the song they've generated without issue 20 times in the past is now so broken. Observable things. Even if it's not the model, then it's the LLM
if they train the LLM and change it, then how it processes prompts will change
doesn't have to mean the model itself has changed, but the LLM probably has
anyone know how to fix the problem that suno fail to create any music?
I had that happen a lot a few days ago when extending
Lyrics have been the biggest factor in that recently
I have had it happen intermittently though
or the LLM, is the easier answer
they do need to update the LLM if they have a new model coming out and improve its accuracy etc
cause I do agree with @charred merlin that its very unlikely they are changing the actual generative model. It would have to be the LLM
I am not sure I understand what you are talking about. Can you explain what the LLM does in suno?
large language model, it processes your inputs (tokens) and translates it to information to provide the model with in how to generate what you are asking it for
that's not what I'm talking about
it's how it understands and interprets prompts
someone is confusing the tokenizer with the LLM itself here
Bark doesn't even start with noise from my understanding.
okay
Bark is an LLM
but either way, something is being done to the front end
an LLM trained to output encodec audio tokens
Hmm.. must have read the documentation wrong
I was creating instrumental music, so there is no lyrics to blameπ₯²
I'm not trying to be a know it all, I want to know where I am mistaken. From documentation Bark is a transformer a unique tokenizer that makes text to audio conversions.
Bark is fully generative text-to-audio model devolved for research and demo purposes. It follows a GPT style architecture similar to AudioLM and Vall-E and a quantized Audio representation from EnCodec. It is not a conventional TTS model, but instead a fully generative text-to-audio model capable of deviating in unexpected ways from any given script. Different to previous approaches, the input text prompt is converted directly to audio without the intermediate use of phonemes. It can therefore generalize to arbitrary instructions beyond speech such as music lyrics, sound effects or other non-speech sounds.
Not sure if you're being sincere or a sarcastic smartass...Either way, you don't know what prompts I'm even using
So i saw the announcement for v4. I guess my question is what exactly is updated or added to the current model? how is it actually different?
That is from Github
They haven't said anything about it
and v4 was kind of obvious all the time
I hope they hold some keynotes
I didn't...I told you I didn't mean the generative model, but the llm
What the hell
I guess im just asking cause without really any technical notes, its no different than slapping a new number next to the current model. I imagine it probably has more training data at the very least?
I can't deal with passive aggressive bs...
LLM and Transformer do not exclude each other
Oh. OK, I'll look into it
I am moving into the direction of AI as a developer so if I am mistaken I need to know about it...I appreciate input
You can also look into AudioLM i think it's what bark is largely based on with some changes of course
Yes, in the documentation that snippet I posted those are links, I'll do that
I asked ChatGPT and it also said it wasn't so then it must be deep
harmonai still around? ^^
As of now, all spins of Bark, AudioLM are not LLMs I don't know if you care or not it may just be me...lol. Vall-e is the closest to an LLM in the set. Even it isn't actually an LLM. I have to keep all this crap straight
Now I want to ask a developer that understands some places I am gapped.
Oh I know a professor that I did my senior Project with that got her PhD in this stuff. I'll ask her.
I hope that bug get fixed in V4, its so annoying
Bark is sort of an LLM. It's based on NanoGPT, which was an normal LLM, adapter to generate/predict audio tokens for a token based audio codec, instead of output tokens that are converted back into human text. You can even use Bark generatively, like an LLM. It does not work well at all, BUT IT IS VERY VERY FUNNY https://x.com/jonathanfly/status/1650001584485552130
If I've been using the wrong term all this time let me know though. But see above for some more info
can we get refunds for this ? lmao https://suno.com/song/c0a0922b-b2fc-4450-86b6-3511a3dff24e
Epic Fast Rap, Male Vocals,fast,angry,depressed,cinematic,deep,heavy, experimental,fast-paced,fast rap, song. Listen and make your own with Suno.
I did read about the nanoGPT. I am mostly keeping things separated and discerning the difference between semantics and important to acknowledge these things. It's why I emailed my old professor. Bark is another topic. From what I have read, Bark does not generate from Noise like stable diffusion or even some other text to audio methods. I read that it is created in latent space creating as it goes. This is very interesting to me. Thanks.
I had that happen...haha. I think a demon may have taken those credits chief...lol
lol, not the first time its happened, but probably the creepiest sound its given me.
Yea that was creepy
last time it did that i was working on a song called "hell bound" so that somewhat made sense
Yea you just go with it
Yes, most audio models in the past worked differently, often like image models including diffusion. Bark was pretty unusual at the time as one of the few that worked in pure tokens like a language model. There are more models like that now. And to make things even more complicated, there are now diffusion models that generate tokens, even for language models, so it's all kind of getting mixed together lol
Awesome. Thank you
@worn flicker so far this is the fastest i've been able to get it on my writing, still just testing it, lyrics aren't done, but like some of the speeds I achieved here https://suno.com/song/97a12646-b537-4ad6-ac15-85e8e84638c7
Sounds great. I suggest doing a tempo change Syncopated Rhythm at the chorus and the another another with Polyrhythmic Percussion to prompt it back to the off beat that you are at. You can even throw out some steady beat to get those syllables to pop in between beats. Adds a whole rhythmic part that makes the flow sound faster if it hits right.
Also, in the chorus use less commas to get it to go faster and longer
i tried the changes you had, [Tempo Change] [Polyrhythmic Percussion]
Pre-Chorus] and [Tempo Change] [Syncopated Rhythm]
[Chorus] and it would sing polyrmthic and syncopated in the song.
You have to put it on the side of the section title to avoid that, and even then it is still gonna happen around 10% of the time
yeah, still working on chorus, once I got the verse done I just kind of wrote a pretty dirty chorus to test with
ok, ill try that,
Since your style is metal then you want enunciation and that is being done well, good job. I like the voice matching up with the beat. Just like the stutter I suggest using vowel expansion or general word expansion to maybe hold a screamed ending to a Verse.
Going to create, It's my fate, Not a debate, can't violate, or deflate, with your (ha-aaaaaaa-aaattte)
Many times that will make it hold and the next Verse will start while the extended word is still going
You can also try {...} but that in my experience tells suno it can skip the word if it would mix up the flow
yeah, i've been working on the end of the verse, from extending it with more rhymes, to echos, to scream, but really like the beat I got on that one, so may build a persona from it and work from that as I finish it
2 more day though π
For sure, then in that case, it is awesome. The creative process rolls, gotta roll with it, I feel that. You have a sound in your head. Gotta get that out...lol
There is no question that if I ran out of credits I would immediately buy more. Suno is becoming a problem for me...lol.
same here, but i have already spent like 120 this month ontop of my subscription soooooo.... need to chill a bit
Oh dang...yea I ain't done all that...haha
There is no rule against multiple paid Suno accounts. It's way cheaper than the "Extra Credits" if you have a heavy Suno month. They really need toupdate those Extra Credit prices, so if that 120 is all Extra Credits, I really recommend using overlow Suno accounts instead, way more efficient unless you are going just a little bit over your cap!
yeah, i probably will, kinda nuts that 30 a month gets me 10000, but 30 on credits only gets me 4000
That is a great tip. Is it bad we had to be told that?...lol we all should have picked up on these things...haha
Extra credits don't expire, so you're not beholden to credits expiring at the end of your billing cycle, but if you know that you'll use more and think to cancel the subscriptions before the end of the month, it's a better value to get multiple paid accounts πͺ
its just an organizational thing for me, don't want to have to remember what account a specific song is on
speaking or organizational things in suno.. man I want a folder structure to organize my songs..
I actually always subscribe at the 2500 tier. If I have a month where I am busy and not using Suno much, I stay on 2500. But if I have a heavy Suno month, I can upgrade mid month to 10000 for 12500 total that month, for $40. And only rarely have I needed overflow accounts, but I have used one occasionally.
that is a good idea
Yes, a good way to use a fluid subscription
I had a couple of months where i didn't use suno so lost 20,000 credits
I have not been a subscriber long. However, do I seem like an amateur...lol. These things are my jam, if it is something to figure out, then I am on it in my spare time. I love pushing AI to its limits. Like getting SD to generate a person doing a handstand without a Lora
It is annoying to split Suno libraries. But also Suno library is not very usable when you have a lot of songs. I'm sure that is temporary, but right now, almost wish I was more split haha.
π« βοΈ
yeah, folder structures would help, folder for completed, folder for ones your working on, etc.
If you go to FoggyX they have multiple accounts. The second account has the other accounts playlists on it. If I am seeing it right. Only one song on the main page. Either that or they made a way to hide songs from the main screen in playlist which would be a folder structure
i've noticed, you can have a public playlist that will show songs in it that arn't public
For a short period, I used a browser tool that added a unique word to the title of every song, and I would change the word basically once a day, or every sit down work "session" but I didn't keep it up. It worked okay for pulling up "those songs I was working on last week" in the library.
Well maybe that is how
I can see how that would get confusing. However it seems as if sessions are not the primary focus for developers considering everything is SSO, it seems like all that can be outsourced, is, to focus on features.
Actually before Suno had library search, frustrated looking for a song I just couldn't find, I channeled the frustration into a crazy weekend project making a super hacky library search https://github.com/JonathanFly/chirp-studio but it broke every time Suno updated their website, and then Suno did add a library search that at least covered the basics, so it's long dead now.
well now that I think about it, it was before suno had public songs, or searchable style tags. So the main reason I wanted it was to literally search my library for examples of specific style tags to figure out what they were doing, since that was actually IMPOSSIBLE between V1 and whenever Suno launched profiles, since V2 was web only
I didn't even see it until version 3, I was deep in SD and life stuff the other versions. I heard about bark but this stuff is moving so fast it's insane. I never thought we would need so much VRAM not being used for games or hooking for more hashing power
"this stuff is moving so fast it's insane" I feel like the technologial power doubles every six months but that window is getting SHORTER so the power grows exponentially now
like it just builds on its own momentum
is there any way to avoid hissing in the vocals..its too harsh!
The simplest method right now is actually to not worry about until you are done, then Cover you own song to fix the voice. This is pretty reliable, as long as your song is short enough to Cover.
yeah, i was born in 1984.. watched all this shit happen, been pretty incredible. was made fun of as a kid for being on a computer, now the entire world is buried in technology
Cover can recreate a song almost exactly the same, but cleaner voice and audio. Not 100% of the time, sometimes a song just doesn't seem to work out, but often enough that I think it's a good default workflow. It is possible to be really careful and keep voice quality very high all the time, but uses a lot of credits and sweat, and it means you can't just pick the best music or song, always have to pay attention to quality, and keep both high the whole time
Hi, I'm trying to βhumβ a melody and have Suno take that melody and put instruments to it. The problem is that Suno doesn't always βstickβ to the melody, sometimes he can βimproviseβ and I don't want that. Is there any way to force Suno to follow my melody?
Probably yes, but there's no simple answer. Suno is often supposed to take inspiration, usually if it's doesn't change things at all, that is more boring. But if you retry different methods and pick the ones that keep the melody the same, you should be able to get a song eventually.
You could also try uploading a longer version with the melody repeated in it (with audio editor) but that might make Suno too stubborn.
What you want ideally is to associate a text prompt with the melody, just like a chorus in a song, and then be able to prompt that and Suno repeats it more reliably. But this is tricky with uploads and especially if there are no words.
Oh, one more thing to try, maybe try Extending the Upload with a blank style prompt and blank lyrics (use space only).
aah suno why! lol, did good till your started repeating bars
I'm gonna try extending the base audio
Oh, in case this wasn't clear: if you are using multiple short extensions, you may have to Get Whole Song to make sure the melody is actually in the musical memory of Suno, if it happened to not be used in a particular Extension in the time before you Extended it again.
Yeah Suno could make it a lot more obvious for sure. Easy to forget for users who have been around for a long time, since it had a been a thing for so long.
I'm getting more consistent results when I'm using a larger uploaded audio. Now it follows the melody but it improvises the 'connections' between parts and similar things.
So when you Extend from an "Part 3", Suno only has memory of the time from the beginning of part 3, up to your Extend From time. And the maximum memory is 2:00. If you Get Whole Song, all the song parts are in one long clip. This also has a limit of 2:00, but all the parts are together. There is also the Crop feature, so if you Get Whole Song, but now you too much memory, you can Crop it.
I'm only doing part 2 at the moment because the results are sometimes not some consistent since part 2 so
In general, longer audio is "stronger" and less likely to be changed by Suno, and shorter audio is "weaker" and more likely to be changed. Not just Uploads, just anything. So if you find the song too repetitive, not creative, and you want to change the music more, then you want shorter memory.
Yep. It was changing almost all the harmony and the song was pretty different
Now it almost feel like a cover
literally this is what made me stop using 10 free 5-songs a day account and buy pro and i never looked back for a second
this replace section is going to be the end of me-- im a perfectionist and now i cant stop making all these 30-second side quests in the middle where there were parts i didnt like
then its like oh well now i've gotta make the rest of the song that dope
and before i know it ive changed the lyrics 100% and it's a different damn song lmao
lol i totally get that, or burnt so many credits just trying to fix things
dude i just generated like 50 30 second clips and it charged me 5 credits?
if u spendin credits like that, then i cant even imagine how picky you must be! hah!
..or maybe,... i cant imagine how GOOD THOSE SONGS ARE ππ€―π€―
well, mostly reruns on persona or cover trying to get that perfect sound, or the times when it gives me the perfect music, but sings it as female so i cover 20 times trying to get the right male tone
Put βfemale voiceβ on exceptions prompt and it always cover with male voice
i have female voice in exceptions and male vocals in style, and still get female a lot (I write like a chick i guess)
For me , in most of cases, it fixes . I get mad when it does a great couber with female voice LOL
right now im trying to figure out rap timing, burnt through a lot on this for testing, lmao (im normaly rock/metal so this was a fun experiment) fastest so far, https://suno.com/song/3f9efe13-0ff3-4a8b-a6bf-638049339794
Epic cinematic Rap, Aggressive, rap, fast, fast-paced, progressive nu metal, experimental, strings, orchestra, war drums song. Listen and make your own with Suno.
Sometimes suno acts like an undomitable beast
i just watched that video, lol whats with the random numbers in the lyrics? thats not in the song..
Hi
I want to ask you what the song was like at the same time?
what do you mean?
Delete the wrong song and don't want to press manually, I want the same?
check your garbage? top right in the library
oh dude i do metal tooo
except i was actually a drummer for a touring thrash/metalcore band for almost 10 years.
went to berklee college of music, and musician's institute, the whole 9
nice! I jump between metal, nu-metal, and some more acoustic style stuff
How do I want to delete the wrong song and delete it at the same time, not one by one?
What to print on the PC
i do metal and rap, and i have a penchant for the really REALLY over-the-top lyrics like the most brutal hardcore or the funniest wildest shit no one would ever say in a billion years lmfao it's like infant annihilator or attilla
a lot of mine are very specific to me and my life, most are pretty dark, and some way over the top (like the bdsm song I did for a friend)
acoustic stuff? hah! i do some of it too but i always have an edge of comedy to my shit that i make just for me lol
or i do covers.
like this:
I intend to record my own vocals for my songs and release them
epic heavy slam beat-down style deathcore, shred technical virtuoso soloist guitar work, epic solos, leads, harmonies, song. Listen and make your own with Suno.
yes from the library do not press one by one
click 1, hold shift, then click another further down
thats dope if i didn't have a compeltely balls voice i'd do this. im a drummer bro lol
Thanks
nice!, do you ever listen to leo on youtube (frogleap)
of course lmao
my band did awesome metal covers too back in the day
we did beat it before even fall out boy did it. when we released ours it was us and raintime and that was it
back in 07
mine are shit too, but i figure with the help of autotune and ai fixes i can make it work
i still wouldn't have my voice out there even with a vocoder and tuned down 2 full steps because i hate the sound of my own voice lmao you're a brave man sir no fkn thanks
haven't done any covers yet, want to do one of fleetwood mac - chain
oh dude we did beastie boys fight for ur right to party
and beat it...
and more human than human
annnnd... PST 88
you know that one?
haven't listened to pantera in awhile but yeah
it's super easy, i did the arrangements on all of ours. literally just played with distorted guitars, double bass through the whole song, and screamed vox. check, done, send it!! haha
kinda digging I miss blockbuster
omfg i dont even know what i have public ive made like 5000 songs but literally have never showed anyone except like my brother i forgot i wrote that song ahahahahahaha
i was thinking about doing like my opinion on current events but in deathcore songs with a girl singer
like with rap in there
lol
so ive got ones like "dear white people: please stop telling people your opinion on the middle east, you're embarrassing the rest of us"
my heaviest song atm, less singing and more just angry screaming, was just venting when I wrote it, wasn't really even trying to rhyme anything.. https://suno.com/song/13ad2533-428f-4411-904e-0079ee77e333
Male Vocals,deep,dark,Angry,depressing, bipolar,Nuclear,black and white,low tone,nu metal,progressive metal,complex song. Listen and make your own with Suno.
or like when the plot from jaws literally happened in texas--- on the actual 4th of july last year.. and nobody was talking about it like wtf man how is no one putting this together-- and THEN IT HAPPENED AGAIN ON THE SAME DAY
lemme peep it-- and no worries, you aint gotta rhyme metal homie-- the more abstract the metaphor, the better
i've done one political song, trying to avoid it though, world is too polarized atm
that was kinda the point of doing it, like take the thing a lot of us are thinking and just outright say it-- the more ridiculous the better
some one is making woodhouse angry π
well you cant say f***
bro i cant say the president-elects name?hahahaha
oh wow
how big of a jackass do you have to be to have anyone make your name a bad word? ahahahahaha
oh it will
yeah it did it. lol funny
because i dunno how old u are... but u remember back in the 90s how if you pierced only your right ear and were a guy you'd get made fun of that you like other dudes?
well when he got bang banged
i was like oh he pierced his right ear
so made a song like "guys i think i can prove DJT likes dudes"